CommonDataModel Introduction

The Library Simplified circulation manager uses a complex data model to represent the current state of:

Looking at the whole data model at once can be overwhelming, so we'll consider it as a few smaller simpler systems:

These systems overlap around a few key classes, mainly DataSource, Identifier, LicensePool, and Work.

The code for the data model is in the model package of the server_core project.

This data model is common between the circulation manager and the metadata wrangler, although some pieces are exclusively used by one component or the other. For example, only the circulation manager has lanes or patrons, and only the metadata wrangler has integration clients. The library registry component has a separate data model which is similar but much simpler.

For the sake of simplicity, this document will talk about "books", but the rules are the same for audiobooks and other forms of content.

Bibliographic metadata

Bibliographic information is information about books as opposed to the books themselves. A book's title, its cover image, and its ISBN are all bibliographic information--the text of the book is not. Bibliographic information flows into the circulation manager and metadata wrangler from a variety of sources, mainly OPDS feeds and proprietary APIs. We keep track of all this information and where it came from, and when necessary we weigh it, sort it, and boil it down into a small amount of information that can be used by other parts of the system.

DataSource

A DataSource is some external entity that puts data into the system. This data generally falls into two categories:

Some examples of DataSources:

A DataSource may also:

Identifier

An Identifier provides a way to uniquely refer to a particular book. Common types of Identifier include ISBNs and proprietary IDs such as Overdrive or Bibliotheca IDs.

An Identifier may:

Equivalency

An Equivalency is an assertion made by a DataSource that two different Identifiers refer to the same book.

Edition

An Edition is a collection of information about a book from a particular data source. Like most items in the "bibliographic metadata" section, it represents an opinion. If different data sources give conflicting information about a book, that's fine -- everyone has their opinion. When this happens, we create multiple Editions and we sort it out later, when it's time to make the presentation edition.

An Edition:

The contributor subsystem

This system basically tracks who wrote which book. There are two classes in this subsystem: Contributor and Contribution.

Contributor

A Contributor is a human being or a corporate entity who is credited with work on some Edition. The credit itself is kept in a Contribution, which ties a Contributor to an Edition.

A Contributor:

Contribution

A Contribution:

The classification subsystem

This system tracks how a book might be classified in a card catalog or shelved in a bookstore. There are two classes in this subsystem: Subject and Classification.

Subject

A Subject represents a classification that someone might give a book. Subject handles a variety of classification schemes: Dewey Decimal, LLC, LCSH, BISAC, proprietary systems like Overdrive's, and free-form tags, among others. Four pieces of information might be derived from the Subject, and will be stored with the Subject if possible:

Classification

A Classification is someone's opinion that a book should be filed under a certain Subject.

A Classification:

Genre

There are many different data sources which use many different classification schemes for the same books. Rather than expose this chaos to patrons, we have defined about 150 Genres, corresponding to the sections of a large bookstore or branch library: "Romance", "Biography", and so on.

Each Subject may be associated with a Genre. When LicensePools are turned into Works, all the related Classifications are gathered together. We then assign the Work to the Genre that showed up the most.

A Genre may also be associated with one or more Lanes -- this is the primary technique we use when choosing how to show books to patrons.

Measurement

A Measurement is a numeric value associated with an Identifier. It represents some quality that distinguishes one book from others. The most useful measurements are popularity (a popular book is read/accessed/purchased/accessioned more often) and rating (a highly rated book is considered to be of high quality).

The linked resources subsystem

This system keeps track of external resources associated with a book. An "external resource" can be pretty much anything, but these are the most common types of resources we track:

Hyperlink

A Hyperlink represents a connection between an Identifier and a Resource. It contains two extra pieces of information about the link:

Resource

A Resource represents a document found somewhere on the Internet -- probably either a cover image or a free book. It has a url, and that's basically it -- everything about the document itself is kept in Representation.

Representation

A Representation is a local cache of a Resource. It represents our attempt to actually download a Resource and records what happened when we tried. If everything went well, the Representation will contain a file--binary, text, HTML, or image. Otherwise, the Representation will contain information about what went wrong -- maybe the server was down or something.

Circulation managers don't usually create Representations -- they rely on the metadata wrangler to do that.

An image Representation that's a thumbnail of another image Representation is connected to its original through .thumbnail_of.

Putting it all together

Here's how the whole subsystem works together. Let's say one of our data sources that claims the URL http://example.org/covers/my-book.png is a cover image for the ISBN "97812345678". We want to represent this fact in our system.

  1. We'd create an Identifier for the ISBN "97812345678".
  2. We'd create a Resource for http://example.org/covers/my-book.png
  3. We'd create a Hyperlink with the rel "http://opds-spec.org/image", for "cover image". The .data_source of this Hyperlink would be set to the DataSource that made the original claim.
  4. We don't have to actually download http://example.org/covers/my-book.png, but if we do decide to download it, the binary image will be stored as a Representation. If there's a problem and we can't complete the download, that fact will be stored in the Representationinstead.
  5. If we download the image and everything goes well, we may also decide to create a thumbnail out of it. This would be stored as a second Representation, and its .thumbnail_of would point to the original, full-size Representation.

ResourceTransformation

A ResourceTransformation represents a change that was made to one Resource to generate another Resource. Currently it's used in the circulation manager's "cover image upload" feature. You can upload a background image (the original Resource) and paste the title and author onto it (a ResourceTransformation which results in a second Resource).

Theoretically, thumbnailing could also be handled as a ResourceTransformation, but it's probably not worth making this change.

Licensing

Collection

A Collection represents a set of books that are made available through one set of credentials.

One library may have multiple collections from different vendors, and multiple libraries on the same circulation manager may share a collection.

A Collection may have children which are also Collections. We model an Overdrive Advantage account as a child Collection of the main Overdrive Collection.

The books themselves are stored as LicensePools, and the credentials are stored in an ExternalIntegration.

LicensePool

A LicensePool represents an agreement on the part of a book vendor to actually deliver a book to a patron.

a group of licenses granting access to one particular Work.

If a Work is not associated with a LicensePool, patrons will not be able to check it out.

In some cases, usually involving open-access LicensePools, there may be more than one LicensePool associated with the same Work; if this happens, the LicensePool which provides the highest-quality version of the book will take precedence.

Each LicensePool:

DeliveryMechanism and LicensePoolDeliveryMechanism

A DeliveryMechanism describes what format a book is actually available in. There are two parts to a DeliveryMechanism: 1) the DRM scheme implemented by the distributor, if any, and 2) the format of the book (EPUB, PDF, audiobook manifest, Kindle, and so on).

LicencePoolDeliveryMechanism is a three-way join table: a record of a promise by a vendor (identified by a DataSource) to deliver copies of a book (identified by an Identifier) in a specific format (identified by a DeliveryMechanism).

RightsStatus

A RightsStatus represents the terms under which a book is being made available to patrons. The most common varieties of RightsStatus are 1) in copyright, 2) public domain, and 3) a Creative Commons license. "In copyright" implies that a book is being made available to patrons by virtue of a licensing agreement between the library and the vendor. The other RightsStatus values imply that a book is being made available to library patrons on the same terms as it would be to the general public.

Complaint

Patrons may lodge one or more Complaints against a specific LicensePool. Complaints indicate problems with specific books. For example, a Patron can lodge a Complaint stating that a book is incorrectly categorized or described, or that there is a problem with checking it out, reading, or returning it.

CirculationEvent

A CirculationEvent is a record of something happening to a LicensePool. A CirculationEvent happens when an event takes place within the circulation manager (e.g. a work is checked out or placed on hold), or when we notice that an event happened on the distributor's side (such as licenses for a book being added or removed), or when a client app (i.e. a book having been opened).

CirculationEvents are aggregated and used to create library analytics.

Works

A Work represents a book in general, as opposed to one specific edition of a book, or a specific licensing agreement to deliver copies of a book.

A Work:

Custom lists

A CustomList is a list of books, typically grouped by a criterion such as genre, subject, bestseller status, etc., which a librarian has compiled in the admin interface. Each CustomList is associated with, and presented to patrons in the front-end as, one Lane. A CustomList has at least one CustomListEntry, each of which refers to a particular Work.

Libraries

Library

A library represents some organization that serves a distinct set of patrons.

Each Library can have:

* one or more `Collections`.
* one or more `CustomLists`.  
* one or more Lanes, each of which is associated with one CachedFeed.
* one or more Admins.

Admin

Admins are people such as librarians who have access to the admin interface (via accounts in the circulation manager). An Admin is associated with a particular Library through AdminRole. An Admin may have more than one AdminRole. The AdminRoles are:

Lane

A library groups its books together using Lanes. A Lane may group books by any combination of these criteria:

A lane may have many CachedFeeds.

CachedFeed

A CachedFeed is a pregenerated OPDS document that's stored in the database to serve future client requests. If a CachedFeed can be used, it greatly improves patron-visible response time.

Any OPDS feed served by the circulation manager can be cached. This includes the various feeds served as a patron browses a Lane, but it also includes the feeds of books by a given author, books in a given series, and recommendations from sources like NoveList.

Patrons

Patron

A Patron object represents a human being who is a patron of some Library. More precisely, a Patron object represents that person's library card. A human being with cards for multiple libraries will have multiple Patron objects.

Most of the information in the Patron record comes from the library's ILS. The most important pieces of information are the three fields used to uniquely identify a patron:

Loan and Hold

These classes are nearly identical. They represent a patron's relationship with a LicensePool -- either they have the right to read the book right now (Loan) or they're waiting in line for the chance to read it (Hold).

Credential

The Patron object keeps track of unique identifiers that identify a patron to their ILS. Other identifiers are stored in Credential objects.

One major type of credential is the “Identifier for Adobe Account ID purposes”. This is an alias provided to Adobe (through the Short Client Token system) whenever the patron needs to activate a mobile device with their Adobe ID.

Another major type of credential is the “OAuth Token”. This is a temporary token granted by an ebook vendor such as Overdrive. It gives the circulation manager the ability to take action on the patron's behalf, e.g. by borrowing books or placing holds.

A Credential may have associated DRMDeviceIdentifiers. This is used to keep track of the device IDs associated with a patron's Adobe ID. This makes the ACS Device Management Protocol possible.

Annotation

A patron in the process of reading a book has a "last reading position" -- the place where they left off. If a patron closes and reopens a book, they expect the book to open their last reading position, not to the beginning of the book.

An Annotation stores a Patron's last reading position for a LicensePool. If the patron creates bookmarks or highlights text, those are also stored as Annotations.

Annotations are synced between client and server using the Web Annotation Protocol. A patron must opt in before their annotations are synchronized with the circulation manager. A patron's decision to opt-in or not is tracked in the field Patron.synchronize_annotations.

Site configuration

ExternalIntegration

A ConfigurationSetting holds information about an extra piece of site configuration. A ConfigurationSetting may be associated with an ExternalIntegration, a Library, both, or neither.

ConfigurationSetting

An ExternalIntegration contains the configuration for connecting to a third-party API. Commonly used third-party APIs include the metadata wrangler, DataSources that require protocols, authentication services, storage services, and search providers.

Background processes